letters
Nuclear envelope formation by chromatin-mediated
reorganization of the endoplasmic reticulum
Daniel J. Anderson1 and Martin W. Hetzer1,2
The formation of the nuclear envelope (NE) around chromatin
is a major membrane-remodelling event that occurs during
cell division of metazoa. It is unclear whether the nuclear
membrane reforms by the fusion of NE fragments or if it reemerges from an intact tubular network of the endoplasmic
reticulum (ER). Here, we show that NE formation and
expansion requires a tubular ER network and occurs efficiently
in the presence of the membrane fusion inhibitor GTPγS.
Chromatin recruitment of membranes, which is initiated by
tubule-end binding, followed by the formation, expansion and
sealing of flat membrane sheets, is mediated by DNA-binding
proteins residing in the ER. Thus, chromatin plays an active
role in reshaping of the ER during NE formation.
The nuclear envelope (NE) is a double membrane that compartmentalizes
eukaryotic cells into nucleoplasm and cytoplasm. The outer (ONM) and
the inner (INM) nuclear membranes, which are fused at sites of nuclearpore complex (NPC) insertion, form a continuous membrane system
with the endoplasmic reticulum (ER)1–3. In metazoa, the NE undergoes
a cycle of disassembly and reformation during cell division, whereas the
ER network remains intact2–7. Evidence from live-cell imaging suggests
that NE components are redistributed into the ER, thus implying that ER
tubules might be the precursor membrane to rebuild the NE6. However,
the mechanisms that would target the ER to chromatin and reorganize
membrane tubules into the flat NE sheets are unknown.
Here, we studied the reorganization of the ER into the NE on the
surface of chromatin using fractionated Xenopus eggs8, which have been
shown to faithfully recapitulate various steps of nuclear assembly3,9,10.
During cell fractionation into the cytosol and membranes11, the fragile
ER network was fragmented (Fig. 1a), making it difficult to discriminate between the molecular requirements for ER reconstitution and NE
formation. To overcome this limitation, we preformed an intact ER by
incubating cytosol with fluorescently labelled membranes for 20 min
and then added sperm chromatin to initiate nuclear assembly (Fig. 1a).
We found that membrane tubules rapidly covered the chromatin surface (Fig. 1a) and that the chromatin-associated network stained with an
antibody against reticulon 4a (Rtn4a), a characteristic protein of tubular
1
2
ER (see Supplementary Information, Fig. S1a). After 60 min incubation,
intact nuclei with a closed NE formed (Fig. 1a), demonstrating that NE
formation from intact ER can be recapitulated in vitro.
Using this experimental strategy, we first asked whether or not NE
formation is sensitive to the non-hydrolysable nucleotide triphosphate
analogues GTPγS and ATPγS. Whereas GTPγS has been shown to
prevent ER reconstitution in vitro by blocking membrane fusion11,12
(see Supplementary Information, Fig. S1b), ATPγS specifically inhibits ER tubule formation without blocking vesicle fusion11 (Fig. 1c, and
see Supplementary Information, Fig. S1b, c). When sperm chromatin
was added to reconstituted ER, nuclear assembly occurred with similar
efficiency in the absence or presence of 4 mM GTPγS and ATPγS. We
noticed that GTPγS nuclei were smaller due to a block in nuclear transport, which is required for nuclear expansion13 (Fig. 1b, c). Importantly,
the nuclei that were formed in the presence of GTPγS and ATPγS were
intact; that is, they had acquired a closed NE, because they excluded
fluorescently labelled streptavidin, an inert protein that cannot diffuse
across an intact NE14 (Fig. 1c). These results were surprising as it had
been shown that NE formation in vitro is strongly inhibited by GTPγS
and ATPγS11,12,15. Our results indicate that the inhibition observed in
previous studies was due to a block in reconstituting a tubular network
from fragmented ER membranes. Consistent with this idea, we found
that both 4 mM GTPγS and ATPγS efficiently blocked NE formation
when initiated from fragmented membranes, as judged by the failure
to exclude streptavidin (Fig. 1c, and see Supplementary Information,
Fig. S1b). Unfused membrane fragments were visible on the chromatin
surface, which did not stain with wheatgerm agglutinin (WGA), a lectin
that binds to nuclear pores (Fig. 1b). We therefore conclude that NE
formation in vitro requires the generation of an intact ER, which involves
homotypic membrane fusion16. However, the formation of the NE from
an intact ER is insensitive to GTPγS and ATPγS.
It was possible that, in contrast to vesicle fusion, ER tubules/
sheets could still fuse17 in the presence of GTPγS, thus explaining why NE formation from the ER was insensitive to GTPγS. To
test this, we assembled ER — labelled separately either with green
(DiOC16) or red (DiIC18) lipophilic dyes — and combined the two
reactions in the absence or presence of 4 mM GTPγS and ATPγS.
Salk Institute for Biological Studies, Molecular and Cell Biology Laboratory, 10010 N. Torrey Pines Road, La Jolla, CA 92037, USA
Correspondence should be addressed to M.W.H. (e-mail: hetzer@salk.edu)
Received 10 April 2007; accepted 30 July 2007; published online 9 September 2007; DOI: 10.1038/ncb1636
1160
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group
letters
c
50
ATPγS Post ER
0
4 min
GT
tE
R
AT
Pγ
Sp
re
ER
AT
Pγ
Sp
os
tE
R
ATPγS Pre ER
10
R
GTPγS Post ER
20
po
s
GTPγS Pre ER
30
Co
Control
Membrane/DNA
b
40
GT
Pγ
S
NE-60
Pγ
Sp
re
E
NE-1
ntr
ol
Reconstituted ER
Percentage of closed NE
Fragmented ER
20 min
NPC
d
Interphase
Cross-section
Surface
Overlay
Metaphase
Membrane proteins
100
80
60
40
20
0
anti-Rtn2
Pre ER
post ER
anti-Rtn4
gp210
POM121
Sec61β
e
Percentage of closed NE
a
Figure 1 The ER network is required for nuclear-envelope formation.
(a) Fragmented endoplasmic reticulum (ER) membranes were stained
with DiOC6 (green) and imaged at 4 ºC using confocal microscopy. The
ER network was reconstituted by incubating membranes with cytosol for
20 min at 25 ºC. Nuclear envelope (NE) formation was induced by adding
nucleoplasmin-decondensed sperm chromatin24; DNA was stained with
Hoechst 3222 (blue) and imaged live after 1 min (NE-1) and 60 min
(NE-60) incubations. (b) Nuclei were assembled as in a in the absence or
presence of 4 mM GTPγS or ATPγS either before (Pre ER) or after (Post ER)
ER reconstitution for 90 min. Before GTPγS was added, sperm chromatin
was first decondensed in membrane-free cytosol. NPCs were stained with
Alexafluor©568 conjugated WGA (red). (c) NEs were formed as described in
b around sperm chromatin-containing biotinylated histones14. After 4 min
or 20 min, Alexafluor©568 conjugated streptavidin was added, which was
excluded only when a closed NE had formed; n = 60. (d) A total of 10 µM
anti-Rtn2 or anti-Rtn4 was added to nuclear assembly reactions either
before or after ER formation, and the percentage of closed nuclei was
determined as described in c; n = 100. (e) HeLa cells were co-transfected
with histone H2B-tdTomato and GFP-fused Sec61β (aa 1–65), and imaged
live during interphase and mitosis. Additionally, HeLa cells stably expressing
POM121–EGFP3 and gp210–EGFP3 were transfected with tdTomato H2B
and imaged during interphase and mitosis. Scale bars, 10 µm.
The fusion of the two ER populations was specifically blocked by GTPγS
but not ATPγS (see Supplementary Information, Fig. S1c). Thus, under
these experimental conditions, GTPγS acts as a general inhibitor of ER
membrane fusion. Taken together, these results suggest that NE formation can occur either by GTPγS-insensitive fusion events or by fusionindependent reorganization of the ER.
One prediction from the latter hypothesis is that the topology of
the tubular ER network is required for NE formation. Recently, it
was shown that the formation of ER tubules is mediated by Rtn4a18.
Consistent with this report, we found that antibodies against Rtn4a, but
not Rtn2, blocked ER network formation in vitro (see Supplementary
Information, Fig. S1d). We incubated cytosol and membranes in the
presence of anti-Rtn4a or anti-Rtn2, and added sperm chromatin.
Closed NE formation was specifically inhibited by anti-Rtn4a. When
anti-Rtn4 was added to preformed ER, NE formation occurred efficiently (Fig. 1d). Importantly, ER membranes could still fuse in the
presence of anti-Rtn4, resulting in the formation of large vesicles17,19,20
(data not shown). These results indicate that the specific membrane
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group
1161
letters
a
DNA
Membranes
d
Membranes
Membranes/DNA
Time
e
t=0
0.3 s
7.1 s
9.3 s
4.9 s
Membranes
b
Ilustration
x
z
Patch outline
y
x
16 s
66 s
t=0
10 s
30 s
f
Membrane patch area (µm2)
6s
Lateral tethering
Tip binding
c
30
28
26
24
22
20
18
16
14
12
10
0
1000
2000
3000
4000
Time (s)
Figure 2 Chromatin-mediated reshaping of the ER. (a) Biotinylated 10 kb
double-stranded DNA was immobilized on streptavidin-coated cover glass
(schematic, left panel) and visualized by Hoechst 3222 (blue, middle
panel). After the addition of cytosol and membranes, a nuclear envelope
(NE) formed specifically on the two-dimensional (2D) chromatin spot (right
panel). Scale bar, 10 µm. (b) Early chromatin/membrane interactions
were imaged with time-lapse microscopy on 2D chromatin at a rate of
~4 frames/s. An initial interaction occurred via membrane tubule tips
(arrowhead). Tubule ends on the surface of 2D chromatin were imaged in
3D by confocal microscopy (lower right panel). Scale bars, 2 µm.
(c) De-condensed sperm chromatin was spun onto coverslips and stained
with Syto©62 (blue). Preformed endoplasmic reticulum (ER) network was
added and membrane/chromatin interactions were instantaneously imaged.
Cross-sections of the chromatin are shown and arrowheads indicate initial
membrane contacts to chromatin. Scale bar, 5 µm. (d) NIH3T3 cells were
plated onto chambered slides and stained with DiOC6 and Hoechst 3222 in
growth medium. Live mitotic cells were imaged with confocal microscopy
and 3D reconstructions were generated. Arrowheads indicate tubule-tip
binding. NE formation was performed as in a. Confocal images were taken
every 10 s. Membrane sheets were schematically outlined (lower panels).
Scale bar, 5 µm. (e) NE formation was performed as in a. Confocal images
were taken every 10 s. Membrane sheets were outlined (lower panels).
Scale bar, 5 µm. (f) Expansion of membrane sheets was monitored by
measuring the surface area of a single patch every 10 s.
topology of ER tubules is required for NE formation. Additionally, the
rate of NE formation in vitro was increased when reactions were started
from intact ER versus ER fragments (see Supplementary Information,
Fig. S1e). One explanation for this effect is that chromatin recruitment
of NE components, such as the nucleoporin POM121, is more rapid
from an intact ER network than from ER fragments (see Supplementary
Information, Fig. S1f).
Our in vitro studies predict that nuclear membrane proteins of the NE
are dispersed throughout the ER during mitosis in metazoa, similar to
the observed mitotic ER localization of the INM protein lamin B receptor6. To test this, we transfected HeLa cells with GFP-tagged versions of
the ER protein Sec61 (aa 1–65) and the two transmembrane nucleoporins POM121 and gp210. As expected, the two nucleoporins localized
specifically to the NE during interphase, whereas Sec61 localized to the
ER and ONM. During cell division, all three proteins stained the mitotic
ER network (Fig. 1e), supporting the idea that NE formation occurs from
the tubular mitotic ER network both in vitro and in vivo.
To study the mechanism by which the ER is reorganized into the NE,
we wanted to analyse this process using time-lapse microscopy. Highresolution live imaging attempts on sperm chromatin and live cells
failed due to the highly condensed nature of sperm chromatin, the high
density of mitotic ER and technical challenges such as focal drift during image acquisition caused by rapid chromatin decondensation (see
Supplementary Information, Movie 1). To overcome these limitations,
we developed a NE formation assay — based on the finding that functional nuclei can be assembled around heterologous double-stranded
DNA (dsDNA)21. We spotted 10 kb linear DNA, biotinylated on one end,
onto streptavidin-coated glass (Fig. 2a). After addition of cytosol, the
immobilized DNA assembled into a two-dimensional (2D) chromatin
surface with a diameter of 800±200 µm (see Supplementary Information,
1162
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group
letters
a
t=0
b
Control
c
2 min
15 min
Naked DNA
Chromatin
Cellular cross-section
Chromatin surface
z
x
y
x
Figure 3 NE formation is mediated by protein–DNA interactions. (a) The
endoplasmic reticulum (ER) network was formed with heat-inactivated
cytosol and added to immobilized DNA and imaged in real-time every 10 s.
(b) Immobilized DNA was first incubated with buffer (left and middle panels)
or cytosol (right panel), ER network was formed with cytosol (left panel)
or heat-inactivated cytosol (centre and right panel) and added to the DNA
surface. (c) Schematic illustration of nuclear envelope (NE) formation. The
ER network is recruited to chromatin via tubule ends, which anchor tubules
on chromatin. Tubule sliding along these immobilized tubules results in
membrane network coating of chromatin. A cross-section of chromatin,
where ER is drawn in green and DNA in blue, illustrates the tubular network
formation on the surface of chromatin (left and centre columns). Surface
image of chromatin illustrates membrane tubule flattening into the NE (right
column). Scale bars, 5 µm.
Fig. S2a). When fluorescently labelled membranes were added, extensive
membrane sheets formed specifically on the chromatin-covered surfaces
(Fig. 2a). These membrane sheets were composed of two lipid bilayers
into which NPCs had assembled, as confirmed by transmission electron
microscopy (see Supplementary Information, Fig. S2b). Additionally,
NPCs were visualized with fluorescently labelled importin-β, a nuclear
transport receptor that stably associates with NPCs13. Pores also contained Tpr, a NE-specific nucleoporin22, suggesting that the formed
membranes are not annulate lamellae, NPC-containing ER cisternae23
(see Supplementary Information, Fig. S2b).
The ability to form NEs on immobilized 2D chromatin allowed us to
monitor membrane dynamics on chromatin at high resolution using
time-lapse confocal microscopy (see Supplementary Information,
Fig. S2c, Movie 2). When imaging was started immediately after addition
of cytosol containing an intact ER network, we observed bright membrane dots of uniform size stably associating with chromatin (Fig. 2b).
Three-dimensional (3D) reconstruction revealed that the observed
dots did not correspond to vesicles but rather to tubule ends bound
to chromatin (Fig. 2b). These ‘anchored’ tubules either bound laterally
to chromatin or allowed the recruitment of additional ER tubules to
the surface of chromatin, which was a result of tubules rapidly sliding
along each other (Fig. 2b, and see Supplementary Information, Movie 3).
The addition of nocodazole did not inhibit tubule binding, suggesting
that microtubules are not required for NE formation in vitro (data not
shown). Tethered tubules, which were, at first, highly dynamic between
a few stable interaction points, bound laterally along chromatin and were
immobilized on the chromatin surface following further incubation (see
Supplementary Information, Fig. S3b, Movie 2).
As ER tubule-end binding on chromatin has not been reported before,
we wanted to analyse whether or not similar interactions can occur
on sperm chromatin. When cytosol with preformed ER was added to
decondensed sperm chromatin24, we observed the instantaneous tubuleend binding from the surrounding ER (Fig. 2c). Consistent with the
observations on 2D chromatin, these tubules bound laterally and rapidly immobilized the ER network on the chromatin surface (Fig. 2b,
see Supplementary Information, Movie 4). Importantly, a similar membrane conformation was observed in vivo. When ER tubules surrounding
mitotic chromatin of NHI3T3 cells were stained with DiOC6 or HeLa
cells were transfected with Sec61–GFP and imaged by confocal microscopy, single membrane dots were visible on the surface of chromatin in
anaphase — which, after 3D reconstitutions of multiple z-sections, were
identified as tubule ends (Fig. 2d and data not shown). Furthermore,
time-lapse microscopy of Sec61–GFP-expressing HeLa cells revealed
that NE formation was initiated by tubule-end binding on telophase
chromosomes (Fig. S2d, Movie 1). Taken together, these results show
that NE formation is initiated by ER tubule-end binding and subsequent
tethering of the ER network on chromatin in vitro and in vivo.
Once the ER was immobilized on 2D chromatin, flat sheets formed by
lateral growth and expanded steadily over time (Fig. 2e, f). As individual
membrane patches were connected with each other and the surrounding ER via tubules, the flattening process reorganized patches into large
membrane sheets, which eventually sealed into a continuous NE (see
Supplementary Information, Fig. S2e). Importantly, during sheet formation, we did not observe the fusion of tubules, which are immobilized at
this stage, or additional recruitment of tubules. Therefore, we conclude
that NE formation occurs by chromatin-mediated membrane flattening,
which could explain why NE formation occurred efficiently in the presence of GTPγS or ATPγS (Fig. 1b). Taken together, these results suggest
that the reorganization of the ER network is sufficient to form the NE.
However, homotypic fusion events involved in ER maintenance are likely
to be required when tubular connections between sheets break11,16,19.
This raises the question of how flattening of ER tubules into NE
sheets is mediated. In a recent report, many NE proteins that are
dispersed in the ER during mitosis have been shown to possess
DNA-binding activity25. These findings prompted us to test whether
or not NE formation can occur on protein-free immobilized DNA.
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group
1163
letters
a
Control
900 rpm
b
Control
NE-120
900 rpm
NE-180
ER
in
t
ac
t6
0
m
in
NE-60
Cross-sectional area (µm2)
d
c
GTPγS
ER
gm
fra
ER intact 60 min
4500
4000
3500
3000
2500
2000
en
1500
d
te
60
1000
m
in
500
0
NE-60
NE-120
NE-120 ER NE-180 ER
fragmented recovered
Figure 4 NE expansion requires an intact ER network. (a) Endoplasmic
reticulum was preformed for 20 min, then 2 mM GTPγS was added and
the membrane network was incubated unperturbed (control) or fragmented
at 900 rpm for 60 min. (b) Membrane fusion was tested by combining
washed membrane fragments with 5 mM ATP and 5 mM GTP and incubated
unperturbed (control), shaken at 900 rpm, or incubated unperturbed with
the addition of GTPγS for 60 min. (c) Nuclei were assembled for 60 min
(NE-60) and then incubated unperturbed or shaken at 900 rpm for 60 min;
both reactions were then incubated an additional 60 min unperturbed.
Nuclei were fixed and imaged at each step. (d) Nuclear cross-sectional areas
from c were measured to determine the extent of nuclear growth; n < 50.
Scale bars, 10 µm.
When an intact ER network was formed in the absence of cytosol18
and added to a protein-free DNA, NE formation occurred (Fig. 3a, and
see Supplementary Information, Movie 5). We noticed that NE formation was more efficient on protein-free DNA in the absence of cytosol
compared with reactions with cytosol or NE formation on pre-formed
chromatin (Fig. 3b, and see Supplementary Information, Movie 1), suggesting that nucleosomes do not contribute to membrane recruitment in
vitro. Additional evidence for the mechanistic role of DNA binding in ER
reorganization on chromatin was obtained by studying tubule alignment
and sheet formation. When DNA was spotted at lower concentrations
(see Supplementary Information, Fig. S3b), ER tubules failed to immobilize even during prolonged incubations compared with control reactions with higher DNA concentrations (see Supplementary Information,
Movie 6). Furthermore, when excess 1 kbp dsDNA was added after the
ER was bound to chromatin, flat sheet formation was inhibited (data
not shown). ER interactions with chromatin were mediated by proteins,
as proteinase K treatment of membranes abolished DNA binding (see
Supplementary Information, Fig. S3a). Taken together, these results
suggest the following model for NE formation. DNA-binding proteins
residing in tubule ends initially bind to chromatin. As the ER is a threedimensional, highly dynamic network with tubules constantly sliding
along each other19, additional DNA-binding events are sufficient to stabilize tubules and to rapidly cover the chromatin surface with a membrane network. Additional tethering of DNA-binding proteins results
in tubules flattening and thereby the merging of sheets, resulting in a
closed NE (Fig. 3c). As the chromatin-bound sheets are connected via
tubules with each other and the surrounding ER, lipids required for
the NE are not limiting and do not necessitate a separate enzymatic
fusion process. However, membrane fusion is required when ER is
fragmented (for example, during fractionation) or when tubules break
and membrane sheets become disconnected from the ER. Therefore, NE
formation requires chromatin-mediated membrane flattening and active
maintenance of the ER by homotypic membrane fusion.
During the final stages of NE formation, small holes, which are formed by
the boundaries of merging sheets, are sealed (Fig. 3c, and see Supplementary
Information, Fig. S2e). An attractive possibility is that these holes are occupied by forming NPCs. In support of this idea, we found that NPC-free
NEs, which formed around DNA beads in the absence of cytosol, failed to
exclude 70 kD, but not 150 kD dextrans (see Supplementary Information,
Fig. S3c, d). In contrast, NEs assembled in the presence of cytosol contained
NPCs and excluded 70 kD dextrans (data not shown), suggesting that pore
assembly might contribute to the final stages of NE sealing.
Once nuclei are formed, they grow in size and the NE expands13. It
has been suggested that vesicle fusion is involved in this process3,26. To
test whether or not the ER network is required for NE expansion or
whether vesicle fusion can account for the size increase, we mechanically fragmented the ER network by constant shaking (Fig. 4a). Vesicle
fusion was not inhibited under these conditions as large vesicles were
generated, which did not form when GTPγS was added (Fig. 4b). Nuclei
were assembled for 60 min and then incubated for an additional 60 min
in the absence or presence of mechanical stress (Fig. 4c). Although the
nuclei expanded dramatically in the control, physical disruption of the
ER blocked NE growth. When mechanical stress was stopped, the ER
network recovered (Fig. 4c, d) and the NE expanded, showing that the
shearing forces did not inactivate the extracts. The nuclei incubated
in constant mechanical stress imported fluorescently labelled BSANLS, excluded 70 kD dextran and inserted NPCs (see Supplementary
Information, Fig. S4a, b), further strengthening the idea that the observed
block in expansion was due to disruption of the NE–ER connections and
not due to compromised nuclear function or transport13.
1164
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group
letters
We have previously shown that NE expansion is blocked by impairing the function of AAA-ATPase p97, which is required for ER network
maintenance in vitro27. Consistent with these data, we found that NE
formation from pre-formed ER occurs efficiently in the presence of
an inhibitory antibody against Ufd1, an adaptor protein of p97 (data
not shown). However, ER network formation was blocked after brief
mechanical stress in the presence of anti-Ufd1 (see Supplementary
Information, Fig. S4c). Importantly, the addition of anti-Ufd1 did
not block membrane vesicle fusion (see Supplementary Information,
Fig. S4d). When ER reconstitution after mechanical stress was inhibited during NE expansion, nuclei failed to grow (see Supplementary
Information, Fig. S4e). Taken together, these results suggest that NE formation and expansion requires an intact ER, which provides the lipids
that are necessary to accommodate the increase in NE surface area.
In summary, we propose that the intrinsic capacity of the ER to form
sheets18 is utilized and coordinated by chromatin (that is, accessible DNA
in chromatin) to mediate the rapid conversion of the ER network into
NE sheets. DNA-binding proteins, which are dispersed in the ER, target
the membranes to chromatin and mediate NE sheet formation. As DNA
binding is a feature of many NE proteins and depletion of individual
polypeptides failed to block NE formation, this is probably a redundant
process25 (data not shown). Interestingly, the recruitment of the nucleoporins POM121 and Ndc1 correlates with the strength of their basic
DNA-binding domains25 (data not shown). It remains to be determined
whether or not tubule-end binding, alignment and sheet formation is
mediated by multiple ER proteins with different DNA-binding affinities or whether structural elements in the ER lumen participate in NE
membrane flattening.
Our data also suggest that homotypic membrane fusion is not the
principal mechanism of NE formation from an intact ER network, but
is required for ER reconstitution and maintenance. We cannot exclude
the possibility that other membrane structures, such as mitotically
generated COPI vesicles20,28, fuse with the ER and contribute to NE
formation. However, we propose that the formation, expansion and
merging of NE sheets is driven by chromatin-mediated shaping of
the ER network using the intrinsic capacity of the tubular ER to form
sheets17. Therefore, enzymatic machinery might not be required to
reorganize the ER into the NE. In contrast, fusion of separated membrane units (for example, fragmented ER or vesicles) requires fusion
machinery, explaining the GTPγS sensitivity of ER reconstitution and
other fusion processes. Interestingly, in the absence of RanGTP, a
small GTPase required for NE formation12, NE formation from intact
ER occurred efficiently, whereas ER fragments immobilized on chromatin did not fuse12 (data not shown). Similarly, nuclear assembly was
blocked when chromatin decondensation occurred in the presence
of GTPγS, followed by the addition of intact ER (data not shown).
This suggests that Ran, which is also involved in nuclear-pore assembly23, and potentially other GTPases, might be involved in chromatin
organization prior to NE formation.
METHODS
Recombinant proteins and antibodies. Importin-β was purified as described
previously23. Antibodies against Rtn2 and Rtn4 were used at 10 µM for inhibition
and at 0.5 µM for immunofluorescence18.
In vitro nuclear and ER assembly. Xenopus egg extracts and sperm chromatin
were prepared as described previously12. Membranes were stained with either
DiIC18, DiOC16 or DiOC6, as previously described27. Nuclei were formed by
either the classical method of combining fragmented ER membranes, cytosol
and sperm chromatin on ice; or by first reconstituting the ER network by mixing
membranes with cytosol and incubating at room temperature for 10–20 min,
then carefully adding sperm chromatin to the intact ER network. Nuclear growth
was monitored by measuring cross-sectional areas using Adobe® Photoshop®
7.0 and Microsoft® Excel® 2004. Nuclear exclusion experiments were carried
out by first de-condensing sperm chromatin with the addition of heat-inactivated cytosol, along with biotinylated histones. Sperm chromatin containing
incorporated biotinylated histones was then combined with membranes and
cytosol; Alexa Fluor® 568 conjugated streptavadin (Molecular Probes, Carlsbad,
CA) was added at various points during nuclear assembly, and rapidly fixed
and quenched with 4% paraformaldehyde along with biotinylated WGA. ER
was assembled with Xenopus egg extracts as described previously11. Intact ER
network was pipetted with cut-off tips to reduce fragmentation. The ER network
and nuclear-assembly reactions were deliberately sheared by shaking in an orbital
shaker at 900 rpm or pipetting five times. Import of BSA-NLS and exclusion of
dextran were carried out as described previously12.
Real-time imaging of nuclear assembly. Sperm chromatin was first de-condensed with heat-inactivated cytosol. A total of 35 mm culture dishes with a
cover-slip insert (MatTek, Ashland, MD) were pre-coated with poly-l-lysine;
de-condensed chromatin was then spotted on dishes and pelleted at 1000 g
for 5 min. Pelleted chromatin was then stained with Syto® 62 (Molecular
Probes) and time-lapse microscopy was performed at 25 ºC, with frames being
acquired every 2 s. A preformed ER network was then added to capture the
initial interactions of nuclear assembly. Two-dimensional nuclear assembly
was carried out using biotinylated dsDNA, which was prepared as described
previously29. DNA was spotted on streptavadin-coated cover slips (Xenoprobe,
Hawthora, NY) and washed three times with TE. Nuclear-assembly reactions
were performed in slide chambers of 10 µl in volume on which the DNA-spotted cover slips were sealed. Time-lapse images were acquired and processed
as described previously13.
In vivo cellular imaging. NIH3T3 cells were grown asynchronously in DMED
with 10% fetal calf serum at 37 ºC in 10% CO2 in Lab-Tek®II 8-chamber coverglass
slides (Nalge Nunc, Rochester, NY). DiOC16 was added to the growth medium
at 0.5 µg ml–1, along with Hoechst 3222 at 1 µg ml–1. Cells were incubated for
10 min, then mitotic cells were imaged live with SP2 confocal microscope (Leica,
Heidelberg, Germany). Z-stacks were then analysed in 3D using Imaris (Bitplane,
Zurich, Switzerland), in which representative 3D projections were generated.
Note: Supplementary Information is available on the Nature Cell Biology website.
Acknowledgements
Antibodies against Rtn4a and Rtn2 were kindly provided by G. Voeltz and Tpr
antibodies by V. Cordes. The histone H2B-tdTomato construct was a kind gift from
Gray Person. We thank members of our lab for helpful discussions and T. Hunter
for critically reading the manuscript. This work was supported in part by the
National Institutes of Health (GM0739994-01). M. W. H is a Pew Scholar.
Author contributions
Both authors contributed to experimental design and data analysis. D.J.A.
performed all experiments. M.W.H. wrote the paper.
Published online at http://www.nature.com/naturecellbiology/
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. D’Angelo, M. A. & Hetzer, M. W. The role of the nuclear envelope in cellular organization. Cell Mol. Life Sci. 63, 316–332 (2006).
2. Burke, B. & Ellenberg, J. Remodelling the walls of the nucleus. Nature Rev. Mol. Cell
Biol. 3, 487–497 (2002).
3. Hetzer, M., Walther, T. C. & Mattaj, I. W. Pushing the envelope: Structure, function, and
dynamics of the nuclear periphery. Annu. Rev. Cell Dev. Biol. 21, 347–380 (2005).
4. Terasaki, M. Dynamics of the endoplasmic reticulum and golgi apparatus during early
sea urchin development. Mol. Biol. Cell 11, 897–914 (2000).
5. Zaal, K. J. et al. Golgi membranes are absorbed into and reemerge from the ER during
mitosis. Cell 99, 589–601 (1999).
6. Ellenberg, J. et al. Nuclear membrane dynamics and reassembly in living cells: targeting of an inner nuclear membrane protein in interphase and mitosis. J. Cell Biol. 138,
1193–1206 (1997).
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group
1165
letters
7. Porter, K. R. & Machado, R. D. Studies on the endoplasmic reticulum. IV. Its form and
distribution during mitosis in cells of onion root tip. J. Biophys. Biochem. Cytol. 7,
167–180 (1960).
8. Lohka, M. J. & Masui, Y. The germinal vesicle material required for sperm pronuclear
formation is located in the soluble fraction of egg cytoplasm. Exp. Cell Res. 148,
481–491 (1983).
9. Gant, T. M. & Wilson, K. L. Nuclear assembly. Annu. Rev. Cell Dev. Biol. 13, 669–695
(1997).
10.Vasu, S. K. & Forbes, D. J. Nuclear pores and nuclear assembly. Curr. Opin. Cell Biol.
13, 363–375 (2001).
11.Dreier, L. & Rapoport, T. A. In vitro formation of the endoplasmic reticulum occurs independently of microtubules by a controlled fusion reaction. J. Cell Biol. 148, 883–898
(2000).
12.Hetzer, M., Bilbao-Cortes, D., Walther, T. C., Gruss, O. J. & Mattaj, I. W. GTP hydrolysis
by Ran is required for nuclear envelope assembly. Mol. Cell 5, 1013–1024 (2000).
13.D’Angelo, M. A., Anderson, D. J., Richard, E. & Hetzer, M. W. Nuclear pores form de
novo from both sides of the nuclear envelope. Science 312, 440–443 (2006).
14.Franz, C. et al. MEL-28/ELYS is required for the recruitment of nucleoporins to chromatin and postmitotic nuclear pore complex assembly. EMBO Rep. 8, 165–172
(2007).
15.Boman, A. L., Delannoy, M. R. & Wilson, K. L. GTP hydrolysis is required for vesicle fusion during nuclear envelope assembly in vitro. J. Cell Biol. 116, 281–294
(1992).
16.Powell, K. S. & Latterich, M. The making and breaking of the endoplasmic reticulum.
Traffic 1, 689–694 (2000).
17.Shibata, Y., Voeltz, G. K. & Rapoport, T. A. Rough sheets and smooth tubules. Cell 126,
435–439 (2006).
1166
18.Voeltz, G. K., Prinz, W. A., Shibata, Y., Rist, J. M. & Rapoport, T. A. A class of membrane
proteins shaping the tubular endoplasmic reticulum. Cell 124, 573–586 (2006).
19.Voeltz, G. K., Rolls, M. M. & Rapoport, T. A. Structural organization of the endoplasmic
reticulum. EMBO Rep. 3, 944–950 (2002).
20.Cotter, L., Allen, T. D., Kiseleva, E. & Goldberg, M. W. Nuclear membrane disassembly
and rupture. J. Mol. Biol. 369, 683–695 (2007).
21.Forbes, D. J., Kirschner, M. W. & Newport, J. W. Spontaneous formation of nucleus-like
structures around bacteriophage DNA microinjected into Xenopus eggs. Cell 34, 13–23
(1983).
22.Onischenko, E. A., Gubanova, N. V., Kieselbach, T., Kiseleva, E. V. & Hallberg, E.
Annulate lamellae play only a minor role in the storage of excess nucleoporins in
Drosophila embryos. Traffic 5, 152–164 (2004).
23.Walther, T. C. et al. RanGTP mediates nuclear pore complex assembly. Nature 424,
689–694 (2003).
24.Philpott, A. & Leno, G. H. Nucleoplasmin remodels sperm chromatin in Xenopus egg
extracts. Cell 69, 759–767 (1992).
25.Ulbert, S., Platani, M., Boue, S. & Mattaj, I. W. Direct membrane protein-DNA interactions required early in nuclear envelope assembly. J. Cell Biol. 173, 469–476 (2006).
26.Burke, B. The nuclear envelope: filling in gaps. Nature Cell Biol. 3, E273–E274
(2001).
27.Hetzer, M. et al. Distinct AAA-ATPase p97 complexes function in discrete steps of
nuclear assembly. Nature Cell Biol. 3, 1086–1091 (2001).
28.Liu, J., Prunuske, A. J., Fager, A. M. & Ullman, K. S. The COPI complex functions in
nuclear envelope breakdown and is recruited by the nucleoporin Nup153. Dev. Cell 5,
487–498 (2003).
29.Heald, R. et al. Self-organization of microtubules into bipolar spindles around artificial
chromosomes in Xenopus egg extracts. Nature 382, 420–425 (1996).
nature cell biology volume 9 | number 10 | OCTOBER 2007
© 2007 Nature Publishing Group